Dr. Schaffer is a theoretical and experimental neuroscientist who uses both approaches to investigate the function of complete neural circuits. He is driven to understand the underlying computations in complex neural circuits that transform sensory information into intelligent behavior. Dr. Schaffer is currently a postdoctoral fellow in the laboratory of Dr. Richard Axel, where he brings computational approaches to the investigation of neural circuits of Drosophila. Dr. Schaffer graduated from Swarthmore College with honors in 2005, and went on to earn his Ph.D. in neuroscience from Columbia University. As a graduate student working with Dr. Larry Abbott, Dr. Schaffer developed models for how recurrently connected neural networks respond to time-varying stimuli. He believes that neuroscience as a field will increasingly require the study of complex neural networks, necessitating both theoretical and experimental approaches. His career goal is to be able to contribute from both directions, running a lab where theory guides experiments, and the data from these experiments informs theoretical models.
“Transforming a neural representation into a memory”
A fundamental property of nervous systems is that they can learn which experiences are associated with good or bad outcomes. Another way to say that is brains take incoming sensory stimuli and associate it with reward or punishment. But where in the brain does this learning take place? How do organisms determine if a new piece of sensory information will lead to good or bad outcomes? Such learning is crucial for an organism to interpret current evens and anticipate the future. These essential questions are difficult to address in humans, so we turned to the fruit fly, whose brains share a surprising number of features with the human brain, especially at the fundamental level of learning based on sensory experience. The brain area of the fly we study is called the mushroom body, which has a role in processing odors. In fact, this structure is known to be required for the fly to associate novel odors with rewards or punishment. Without the mushroom body, the fly cannot learn. So how does the mushroom body enable the fly to learn? The neurons that make up the mushroom body respond uniquely to each odor the fly detects. However, those responses, while unique, are essentially random—a strategy which at first might seem useless. A consequence of random representation is that odors are not inherently good or bad. It is up to the neurons at the next stage of processing—that is, those neurons that receive input from mushroom body neurons—to learn whether a given odor will result in a reward or in punishment. These same neurons, it turns out, also receive input from neurons programmed to respond to innately attractive or aversive stimuli, such as food or an electric shock. This convergence of information onto these neurons suggests that they are the sites of learning. The next time a fly smells an odor that previously led to an electric shock, it will avoid that odor. In our experiments, we will not present and odor to the fly. Instead, we will artificially activate random groups of those mushroom body neurons that would normally respond to an odor. We will then pair this pattern of activation with a stimulus, such as a shock. In this way, we can test whether or not the fly can learn to seek or avoid a stimulus associated with any arbitrary pattern of neural activity. Finally, once the animal has learned that a pattern of neural activation is associated with a reward or punishment, we can then use this setup to ask how neurons that receive input from the odor-responding neurons represent the reward or punishment association and ultimately drive the fly’s actions. Collectively, our experiments will shed light on the fundamental question of how neurons take patterns of activity and learn to associate them with rewards or punishments.